A mandatory, systematic 4-phase debugging framework designed to identify and fix root causes rather than just treating symptoms.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install taku-debug
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install taku-debug using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is taku-debug?
taku-debug is a specialized skill for AI agents that enforces a disciplined approach to troubleshooting. It operates on an Iron Law: no code fixes are allowed until a systematic investigation is complete. By merging evidence-based investigation with pattern matching, this tool ensures that developers don't just patch surface-level errors but actually resolve the underlying architectural or logical failures.
Built for high-stakes environments where random fixes can introduce new regressions, this skill within the Openclaw Skills ecosystem guides the agent through evidence collection, pattern recognition, and hypothesis testing. It prevents the common failure mode of jumping to conclusions, ensuring every fix is backed by a verified root cause and a regression test.
taku-debug Use Cases
- When the verify phase fails and the failure path is not understood.
- Handling user-reported crashes, 500 errors, or unexpected null returns.
- Investigating intermittent regressions that only appear in specific environments like CI or staging.
- Explaining the technical why behind a failure before proposing a code change.
- Tracing silent failures or data corruption where the source is unknown.
How taku-debug Works
- Phase 1: Investigate: Gather raw evidence by reading full stack traces, reproducing the bug consistently, and tracing data flow backward to its origin.
- Phase 2: Pattern: Match the observed behavior against known failure patterns such as race conditions, null propagation, or config drift.
- Phase 3: Hypothesis: Rank potential causes by probability and test them minimally one at a time to isolate the specific variable at fault.
- Phase 4: Implement: Create a failing regression test first, then apply the minimal fix to the root cause and verify that the entire suite passes.
taku-debug Setup
To integrate this debugging workflow into your environment, ensure the skill is registered within your agent configuration. You can check recent changes during investigation using standard git commands:
# Check the last 20 changes for affected files
git log --oneline -20 -- <affected-files>
# View the diff of the most recent change
git diff HEAD~1 -- <affected-files>
taku-debug Data Schema & Taxonomy
The skill generates a structured Debug Report to document the investigation and its outcome:
| Field | Description |
|---|---|
| Symptom | A description of the observed broken behavior. |
| Root Cause | The specific file, function, or condition found to be at fault. |
| Hypotheses | A log of confirmed and denied theories during the process. |
| Fix | Minimal code change with specific file and line references. |
| Evidence | Proof of the failing (red) and passing (green) test cycles. |
| Status | Final state: DONE, DONE_WITH_CONCERNS, or BLOCKED. |
taku-debug Advanced Features
- Iron Law Enforcement: Prevents the agent from writing fix code until investigation phases are logged.
- 3-Strike Rule: Automatically triggers an escalation or instrumentation request if three hypotheses fail.
- Blast Radius Check: Monitors the scope of changes; fixes touching more than 5 files require manual justification.
- Learning Integration: Checks .taku/learnings for project-specific historical bug patterns.
- Multi-Component Diagnostics: Guidelines for adding diagnostic logging across API, service, and database boundaries.
SKILL.md
Loading
Related Openclaw Skills
Free to No Watermark
A professional AI-driven video processing skill that removes watermarks and cleans raw footage via cloud GPU rendering.
Video To GIF
A high-performance utility to transform video files into optimized animated GIFs or WebP images using ffmpeg and smart palette generation.
Free Editor No
A cloud-powered AI video editing skill that transforms raw footage into polished content via simple natural language commands.
Free No Generator
A high-speed AI video editing skill that processes existing MP4 files into professional 1080p content without consuming AI generation credits.
Free Generator No
A powerful AI video generation skill that transforms text prompts into professional 1080p video clips without the need for paid subscriptions.
Taku Review
An automated code review and delivery gate skill that ensures implementation matches intent while catching critical security and logic errors.
Featured*
